| blob | 3ab6db35dfb0d36664d28a31e1e5301237acf3e9 |
1 ======================================================
2 Kaleidoscope: Conclusion and other useful LLVM tidbits
3 ======================================================
5 .. contents::
6 :local:
8 Tutorial Conclusion
9 ===================
11 Welcome to the final chapter of the "`Implementing a language with
12 LLVM <index.html>`_" tutorial. In the course of this tutorial, we have
13 grown our little Kaleidoscope language from being a useless toy, to
14 being a semi-interesting (but probably still useless) toy. :)
16 It is interesting to see how far we've come, and how little code it has
17 taken. We built the entire lexer, parser, AST, code generator, and an
18 interactive run-loop (with a JIT!) by-hand in under 700 lines of
19 (non-comment/non-blank) code.
21 Our little language supports a couple of interesting features: it
22 supports user defined binary and unary operators, it uses JIT
23 compilation for immediate evaluation, and it supports a few control flow
24 constructs with SSA construction.
26 Part of the idea of this tutorial was to show you how easy and fun it
27 can be to define, build, and play with languages. Building a compiler
28 need not be a scary or mystical process! Now that you've seen some of
29 the basics, I strongly encourage you to take the code and hack on it.
30 For example, try adding:
32 - **global variables** - While global variables have questional value
33 in modern software engineering, they are often useful when putting
34 together quick little hacks like the Kaleidoscope compiler itself.
35 Fortunately, our current setup makes it very easy to add global
36 variables: just have value lookup check to see if an unresolved
37 variable is in the global variable symbol table before rejecting it.
38 To create a new global variable, make an instance of the LLVM
39 ``GlobalVariable`` class.
40 - **typed variables** - Kaleidoscope currently only supports variables
41 of type double. This gives the language a very nice elegance, because
42 only supporting one type means that you never have to specify types.
43 Different languages have different ways of handling this. The easiest
44 way is to require the user to specify types for every variable
45 definition, and record the type of the variable in the symbol table
46 along with its Value\*.
47 - **arrays, structs, vectors, etc** - Once you add types, you can start
48 extending the type system in all sorts of interesting ways. Simple
49 arrays are very easy and are quite useful for many different
50 applications. Adding them is mostly an exercise in learning how the
51 LLVM `getelementptr <../LangRef.html#getelementptr-instruction>`_ instruction
52 works: it is so nifty/unconventional, it `has its own
53 FAQ <../GetElementPtr.html>`_! If you add support for recursive types
54 (e.g. linked lists), make sure to read the `section in the LLVM
55 Programmer's Manual <../ProgrammersManual.html#TypeResolve>`_ that
56 describes how to construct them.
57 - **standard runtime** - Our current language allows the user to access
58 arbitrary external functions, and we use it for things like "printd"
59 and "putchard". As you extend the language to add higher-level
60 constructs, often these constructs make the most sense if they are
61 lowered to calls into a language-supplied runtime. For example, if
62 you add hash tables to the language, it would probably make sense to
63 add the routines to a runtime, instead of inlining them all the way.
64 - **memory management** - Currently we can only access the stack in
65 Kaleidoscope. It would also be useful to be able to allocate heap
66 memory, either with calls to the standard libc malloc/free interface
67 or with a garbage collector. If you would like to use garbage
68 collection, note that LLVM fully supports `Accurate Garbage
69 Collection <../GarbageCollection.html>`_ including algorithms that
70 move objects and need to scan/update the stack.
71 - **debugger support** - LLVM supports generation of `DWARF Debug
72 info <../SourceLevelDebugging.html>`_ which is understood by common
73 debuggers like GDB. Adding support for debug info is fairly
74 straightforward. The best way to understand it is to compile some
75 C/C++ code with "``clang -g -O0``" and taking a look at what it
76 produces.
77 - **exception handling support** - LLVM supports generation of `zero
78 cost exceptions <../ExceptionHandling.html>`_ which interoperate with
79 code compiled in other languages. You could also generate code by
80 implicitly making every function return an error value and checking
81 it. You could also make explicit use of setjmp/longjmp. There are
82 many different ways to go here.
83 - **object orientation, generics, database access, complex numbers,
84 geometric programming, ...** - Really, there is no end of crazy
85 features that you can add to the language.
86 - **unusual domains** - We've been talking about applying LLVM to a
87 domain that many people are interested in: building a compiler for a
88 specific language. However, there are many other domains that can use
89 compiler technology that are not typically considered. For example,
90 LLVM has been used to implement OpenGL graphics acceleration,
91 translate C++ code to ActionScript, and many other cute and clever
92 things. Maybe you will be the first to JIT compile a regular
93 expression interpreter into native code with LLVM?
95 Have fun - try doing something crazy and unusual. Building a language
96 like everyone else always has, is much less fun than trying something a
97 little crazy or off the wall and seeing how it turns out. If you get
98 stuck or want to talk about it, feel free to email the `llvm-dev mailing
99 list <http://lists.llvm.org/mailman/listinfo/llvm-dev>`_: it has lots
100 of people who are interested in languages and are often willing to help
101 out.
103 Before we end this tutorial, I want to talk about some "tips and tricks"
104 for generating LLVM IR. These are some of the more subtle things that
105 may not be obvious, but are very useful if you want to take advantage of
106 LLVM's capabilities.
108 Properties of the LLVM IR
109 =========================
111 We have a couple common questions about code in the LLVM IR form - lets
112 just get these out of the way right now, shall we?
114 Target Independence
115 -------------------
117 Kaleidoscope is an example of a "portable language": any program written
118 in Kaleidoscope will work the same way on any target that it runs on.
119 Many other languages have this property, e.g. lisp, java, haskell,
120 javascript, python, etc (note that while these languages are portable,
121 not all their libraries are).
123 One nice aspect of LLVM is that it is often capable of preserving target
124 independence in the IR: you can take the LLVM IR for a
125 Kaleidoscope-compiled program and run it on any target that LLVM
126 supports, even emitting C code and compiling that on targets that LLVM
127 doesn't support natively. You can trivially tell that the Kaleidoscope
128 compiler generates target-independent code because it never queries for
129 any target-specific information when generating code.
131 The fact that LLVM provides a compact, target-independent,
132 representation for code gets a lot of people excited. Unfortunately,
133 these people are usually thinking about C or a language from the C
134 family when they are asking questions about language portability. I say
135 "unfortunately", because there is really no way to make (fully general)
136 C code portable, other than shipping the source code around (and of
137 course, C source code is not actually portable in general either - ever
138 port a really old application from 32- to 64-bits?).
140 The problem with C (again, in its full generality) is that it is heavily
141 laden with target specific assumptions. As one simple example, the
142 preprocessor often destructively removes target-independence from the
143 code when it processes the input text:
145 .. code-block:: c
147 #ifdef __i386__
148 int X = 1;
149 #else
150 int X = 42;
151 #endif
153 While it is possible to engineer more and more complex solutions to
154 problems like this, it cannot be solved in full generality in a way that
155 is better than shipping the actual source code.
157 That said, there are interesting subsets of C that can be made portable.
158 If you are willing to fix primitive types to a fixed size (say int =
159 32-bits, and long = 64-bits), don't care about ABI compatibility with
160 existing binaries, and are willing to give up some other minor features,
161 you can have portable code. This can make sense for specialized domains
162 such as an in-kernel language.
164 Safety Guarantees
165 -----------------
167 Many of the languages above are also "safe" languages: it is impossible
168 for a program written in Java to corrupt its address space and crash the
169 process (assuming the JVM has no bugs). Safety is an interesting
170 property that requires a combination of language design, runtime
171 support, and often operating system support.
173 It is certainly possible to implement a safe language in LLVM, but LLVM
174 IR does not itself guarantee safety. The LLVM IR allows unsafe pointer
175 casts, use after free bugs, buffer over-runs, and a variety of other
176 problems. Safety needs to be implemented as a layer on top of LLVM and,
177 conveniently, several groups have investigated this. Ask on the `llvm-dev
178 mailing list <http://lists.llvm.org/mailman/listinfo/llvm-dev>`_ if
179 you are interested in more details.
181 Language-Specific Optimizations
182 -------------------------------
184 One thing about LLVM that turns off many people is that it does not
185 solve all the world's problems in one system (sorry 'world hunger',
186 someone else will have to solve you some other day). One specific
187 complaint is that people perceive LLVM as being incapable of performing
188 high-level language-specific optimization: LLVM "loses too much
189 information".
191 Unfortunately, this is really not the place to give you a full and
192 unified version of "Chris Lattner's theory of compiler design". Instead,
193 I'll make a few observations:
195 First, you're right that LLVM does lose information. For example, as of
196 this writing, there is no way to distinguish in the LLVM IR whether an
197 SSA-value came from a C "int" or a C "long" on an ILP32 machine (other
198 than debug info). Both get compiled down to an 'i32' value and the
199 information about what it came from is lost. The more general issue
200 here, is that the LLVM type system uses "structural equivalence" instead
201 of "name equivalence". Another place this surprises people is if you
202 have two types in a high-level language that have the same structure
203 (e.g. two different structs that have a single int field): these types
204 will compile down into a single LLVM type and it will be impossible to
205 tell what it came from.
207 Second, while LLVM does lose information, LLVM is not a fixed target: we
208 continue to enhance and improve it in many different ways. In addition
209 to adding new features (LLVM did not always support exceptions or debug
210 info), we also extend the IR to capture important information for
211 optimization (e.g. whether an argument is sign or zero extended,
212 information about pointers aliasing, etc). Many of the enhancements are
213 user-driven: people want LLVM to include some specific feature, so they
214 go ahead and extend it.
216 Third, it is *possible and easy* to add language-specific optimizations,
217 and you have a number of choices in how to do it. As one trivial
218 example, it is easy to add language-specific optimization passes that
219 "know" things about code compiled for a language. In the case of the C
220 family, there is an optimization pass that "knows" about the standard C
221 library functions. If you call "exit(0)" in main(), it knows that it is
222 safe to optimize that into "return 0;" because C specifies what the
223 'exit' function does.
225 In addition to simple library knowledge, it is possible to embed a
226 variety of other language-specific information into the LLVM IR. If you
227 have a specific need and run into a wall, please bring the topic up on
228 the llvm-dev list. At the very worst, you can always treat LLVM as if it
229 were a "dumb code generator" and implement the high-level optimizations
230 you desire in your front-end, on the language-specific AST.
232 Tips and Tricks
233 ===============
235 There is a variety of useful tips and tricks that you come to know after
236 working on/with LLVM that aren't obvious at first glance. Instead of
237 letting everyone rediscover them, this section talks about some of these
238 issues.
240 Implementing portable offsetof/sizeof
241 -------------------------------------
243 One interesting thing that comes up, if you are trying to keep the code
244 generated by your compiler "target independent", is that you often need
245 to know the size of some LLVM type or the offset of some field in an
246 llvm structure. For example, you might need to pass the size of a type
247 into a function that allocates memory.
249 Unfortunately, this can vary widely across targets: for example the
250 width of a pointer is trivially target-specific. However, there is a
251 `clever way to use the getelementptr
252 instruction <http://nondot.org/sabre/LLVMNotes/SizeOf-OffsetOf-VariableSizedStructs.txt>`_
253 that allows you to compute this in a portable way.
255 Garbage Collected Stack Frames
256 ------------------------------
258 Some languages want to explicitly manage their stack frames, often so
259 that they are garbage collected or to allow easy implementation of
260 closures. There are often better ways to implement these features than
261 explicit stack frames, but `LLVM does support
262 them, <http://nondot.org/sabre/LLVMNotes/ExplicitlyManagedStackFrames.txt>`_
263 if you want. It requires your front-end to convert the code into
264 `Continuation Passing
265 Style <http://en.wikipedia.org/wiki/Continuation-passing_style>`_ and
266 the use of tail calls (which LLVM also supports).